TELEVISIÓN, PADRES, HIJOS Y EDUCACIÓN
2.3. Criterios para ver televisión con los niños
2.3.2.1 Definition
The Pacific Northwest National Laboratory (PNNL) introduces the edge computing [56] as “an approach to move the applications, data and services to logical extremes of the network and it allows information and analytics to occur at the source of the data”.
The Edge Computing Consortium (ECC) defines the edge computing [57] as an open platform deployed on the edge of the network that is close to the source of the data, and provides intelligent services to meet the requirements of real-time processing, data optimization, security and privacy by mobile edge network infrastructure [58].
OpenEdge Computing defines “edge computing as computation done at the edge of the network through small data centers that are close to users”[59]
.
“The original vision for edge computing is to provide compute and storage resources close to the user in open standards and ubiquitous manner” [60].2.3.2.2 Where is Edge?
As discussed above, the generated data by end devices is computed at the edge or close to the edge of the network in edge paradigm. Here, the core networks equivalent is the edge of the network where end devices directly generate the data from surroundings.
Edge computing (EC) adds a new tier of connectivity at the edge of the network between centralized cloud and end-devices. Edge computing enhance the cloud services efficiently such as computations, processing and management close, up to one hop away from IoT devices in the local network such as the WiFi access points or gateways Instead of depending on the cloud hundreds of centralized cloud data centres [59][60]. It allows the services to utilize the devices available in the vicinity e.g. by offering real-time communication, high data rate and ultra-low latency and also has the capacity to control and limit the private user data.
32
European telecommunication Standard Institute (ETSI) proposed Multi-access Edge Computing (MEC) a standard solution for forth coming 5G networks. Instead of transferring all the data from the end devices to the centralized cloud, MEC offload the data to the edge of the network for processing and data storage from mobile and IoT devices [61][62][63][64][65]. Figure 12 shows the proximity of end devices in the edge layer.
Figure 12. Edge/Fog devices placement in the network system [66].
Comparing with edge computing, MCC (Mobile Cloud Computing) also move the capabilities of the mobile devices and enhance management, storage, computing of end devices generated data. Edge computing is dissimilar with MCC, as it provide computing, processing and analysing at the edge of the network close to the end devices. Edge paradigm offers pre- processing, data filtering IoT data via cloud services installed close to the IoT devices by integrating IoT devices with cloud [59].
2.3.2.3 Edge vs Fog
Fog jointly works with the cloud, while edge is defined by the exclusion of cloud [54]. Although, the term used as FC is somewhat close to edge computing. With various overlapping definitions described in the literature for both Fog and EC computing, it is still unclear to differentiate between them [61][67][68]. OpenFog Consortium also distinguish that fog computing works in a hieratically manner and it offers storage , offloading , processing ,
computing control anywhere between cloud and things whereas edge computing appears to be restricted for computing to the edge of the network [69].
Chiang et al. [70] fog computing comprises cloud, core, metro, edge, clients, and things. Fog architecture distributes orchestration, managing and securing the resources and functions in the cloud, anywhere cloud-to-end-devices continuum, and support end-to-end services and applications on the things. Instead of treating edge network as isolated computing platforms, it pursue seamless computing services from cloud to the end-devices.
Harjula et el. assumes fog computing is mainly used to provide platform for services which is above of edge network and local end devices network while edge computing primarily refers to the operational edge network/infrastructure. For better system performance at the edge of the network and minimize the core and cloud consumption/load and increase the network durability, fog include pre-process , cache and analytics the IoT devices generated data before send to the cloud [61][68].
2.3.2.4 Benefits of EC
Edge computing is used to reduce the core network load and is not used to eliminate cloud computing, but it a new addition layer in the network system for processing. Because of its cutting-edge software capabilities various business services have transitioned from cloud to edge, there are various advantages of using edge-computing paradigm for IoT solutions. Few of these are discussed as follows [72].
Trust: With edge computing, the data privacy of local user is safer than cloud and fog computing as the user data remains in the lowest layer and it is easy to manage and control from intruders [72].
Proximity: Communicating and sharing information between the close nodes is more effective than using distant traditional cloud servers. In 19080s and 1990s, peer-to-peer networks gained popularity in this context [72].
Intelligence: As mentioned above, new edge devices have more power capacity and can offer more tasks/instruction to be processed on the edge. This opens the door to automated decision making on the edge, such as distributed crowd-sensing applications or agents that can respond to incoming information flows [72].
Control: The application is controlled and managed in the devices at the edge. Such devices can allocate or delegate to other peers or to the cloud computing, scheduling or storage [72][73].
Latency: In EC, the response time in computation services is counted in milliseconds and supports various SaaS schemes. EC can perform data analytics, predictive analysis and virtualization on edge servers. Relying on its lower latency, EC enables ubiquitous computing in smart applications, where the user can interact with the system in real time and have a better Quality of Experience (QoE). Smart applications, which requires low latency where a local user can communicate with the system in real time and have good Quality of Service (QoS) in edge computing (EC) [71].
34
Human: User’ sensitive information should be computed and storage close to them in order to keep humans in charge of their knowledge [72].
Bandwidth and Scalability: By 2020, 50 billion end-devices produce a huge amount of data, which, send to the cloud using MANET applications such as video streaming, online games, e-commerce, etc. Hence, increase the overall load on the network. EC enables the processing and computing at the edge server can reduce the amount of data to the upper layers of the network, improve the energy efficiency, and reduce the bandwidth utilization in MANET applications. In addition, EC offers low latency for critical applications which requires a prompt reaction for lifesaving events such as in VANET and IoV’s. Therefore, the transport network prevents from frequent accidents and it is a strong bend towards EC paradigm for many smart city projects across the globe such as e-health, smart transport [71].
Cost Effective: Centralized cloud servers are cheap for data storage but expensive to get it out. It is reverse in edge.
2.3.2.5 Limitations
In edge computing, processing nodes are geographically distributed. In fact, edge-based services have to cope with different aspects of constrained environment. This section identifies and address the potential issues in the context of edge computing.
Security and privacy: The most critical services such as data security protection and privacy should be provided at the networks edge. For Example, in a smart home, private data can be analyzed easily through sensor usage data. Intruder/hacker can easily speculate whether or not the house is vacant through the usage of electricity or water usage reading. In this context, it is a problem how to provide service without harming the user’s privacy in CC. To keep the data in the edge network for computing, which implies at home, may be an optimal solution to protect the data security and the privacy. The traditional security and privacy mechanism used in CC is not a better solution for edge paradigm. There should be some new security algorithm introduced by the researchers according to the capacity-constraint edge devices. To offer protection against data security issues, it is important to model a lightweight authentication mechanism wherein EC servers authenticate the IoT devices without a time delay. In order to handle this issue, there should be a reliable trust management system incorporate in the edge servers which is capable enough to manage the end nodes and edge servers [73]. Trust issue: As edge servers are geographically distributed over the network, the trust estimate from one EC server cannot headlong the confidence to the other EC servers. In the distributed networks such as VANETs and MANETs, end-devices are mobile and requires time-to-time authentication. An appropriate trust mechanism needs to be deployed in the EC servers, which are capable enough to manage the trust both from servers and from end nodes [74].
Programmability: Users program their code and deploy it on the centralized cloud server. In the cloud, service provider is in charge to decide on which computing device this computation will conduct. Customer/users have limited information of how the application runs, as the cloud infrastructure is transparent to the user. The code is usually written in one programming language and optimized for different target platforms, as the application only operates in a cloud. However, computation is offloaded from cloud in the edge computing, and the edge nodes comprised heterogeneous platforms. It is very difficult for programmers to write an application and deploy in the edge computing as edge device manufacture varies from each other [73].
Naming: As the number of end-devices are large and there are many applications which, runs the services according to the application’s requirement on the edge nodes, need for naming scheme in the edge computing like all computer systems for programming, addressing, and data communication is very important [73]. Hence, an effective naming scheme for the edge computing model is yet to be developed and standardized [75]. To link with the heterogeneous objects, typically edge operators require learning specific communication and network protocols within their network system. The main aim of the naming scheme is to cope the dynamic network topology, end-devices mobility, security and privacy. Most of the current networks are well managed using traditional Doman Name Service (DNS) and uniform resource identifier. Due to dynamic edge network and mobility of end-devices, this scheme is not flexible handle these network.